What is a dungeon? A series of events with a predictable progression. Typically a fun-house of horror and pain. But is the dungeon aesthetic inseparable from the experience? I think plenty of people have proven that it isn't, and this got me thinking: my players are aware of an item that cannot be destroyed because it is part of a god, they hate said item so much that they are planning to go and kill the god. Specifically, Duven'Ku and the Necklace of the Sleeping Queen.
Of all the horrible things I've done to my players, Death Love Doom struck a cord, thanks +James Raggi. One player retired his character to a monastic life and took a break from gaming because of that session, which I consider a great success. The surviving character hasn't forgotten the almost-TPK that cracking open the necklace caused, and has left a search for the creature on a low simmer ever since.
So, killing gods it is. A dead god of the dead, in a dead city in a dead land. We know his city is at the bottom of a canyon, we know it's dormant or at least incapable of projecting its malignant nastiness any more. Let's assume it has been sealed up as a result of a bitter war or some sort, a war that Duven'Ku got the short end of. The city is sealed, to a degree, and probably large since Duven'Ku was a three-in-one religion/magic-tradition/state. Let's also assume that Duven'Ku resides there, Cthulhu style.
|Replace the sand with snow and you're there|
The issue with a city is that it is far easier to meander about than in an underground complex. A simple approach would be to roughly map the city, set up distinct districts, and then maybe create an enormous random table for generating buildings within that area. A random building for each "room", which you fill in on the map "here is event #24, it will always be here from now on". Each area will have its own flavour and grow over time as they are explored.
I really like this idea myself, but the execution is going to be another thing altogether. The map has been a right pickle so far, my room is full of abandoned attempts, and writing hundreds of interesting encounters is going to be a stretch to say the least.
This will be an ongoing project, since my players are set on going there eventually anyway. Hopefully it will develop into something usable and scalable.